Re: [PATCH] TTY: pty, fix pty counting

[Ilya Zykov]

Jiri Slaby wrote:
> On 10/23/2011 11:01 PM, Ilya Zykov wrote:
>> New version for commit: 24d406a6bf736f7aebdc8fa0f0ec86e0890c6d24
> 
> Although it will work, as ptms are not allowed to be reopen, it doesn't
> look correct. We should decrement the count in ->remove, because we are
> incrementing in install.
> 
> Now, when I understand ptm+devpts layer a bit more, instead of the
> current hackish approach introduced by 24d406a6b (TTY: pty, fix pty
> counting), I think we may introduce a ->remove hook specific to ptms. In
> that one we could decrement the count and don't bother with the
> pty_count macros anymore. Right?
> 
> BTW you cannot remove ->remove hook of pty layer. It would cause an OOPS
> because driver->ttys is not allocated for ptys.
> 
>> diff -uprN a/drivers/tty/pty.c b/drivers/tty/pty.c
>> --- a/drivers/tty/pty.c	2011-05-19 08:06:34.000000000 +0400
>> +++ b/drivers/tty/pty.c	2011-10-23 18:01:20.000000000 +0400
>> @@ -36,13 +36,15 @@
>>  static struct tty_driver *ptm_driver;
>>  static struct tty_driver *pts_driver;
>>  #endif
>> +static int pty_count;
>>  
>>  static void pty_close(struct tty_struct *tty, struct file *filp)
>>  {
>>  	BUG_ON(!tty);
>> -	if (tty->driver->subtype == PTY_TYPE_MASTER)
>> +	if (tty->driver->subtype == PTY_TYPE_MASTER) {
>>  		WARN_ON(tty->count > 1);
>> -	else {
>> +		pty_count--;
>> +	} else {
>>  		if (tty->count > 2)
>>  			return;
>>  	}
>> @@ -446,7 +448,6 @@ static inline void legacy_pty_init(void)
>>  int pty_limit = NR_UNIX98_PTY_DEFAULT;
>>  static int pty_limit_min;
>>  static int pty_limit_max = NR_UNIX98_PTY_MAX;
>> -static int pty_count;
>>  
>>  static struct cdev ptmx_cdev;
>>  
>> @@ -599,15 +600,9 @@ free_mem_out:
>>  	return -ENOMEM;
>>  }
>>  
>> -static void pty_unix98_remove(struct tty_driver *driver, struct tty_struct *tty)
>> -{
>> -	pty_count--;
>> -}
>> -
>>  static const struct tty_operations ptm_unix98_ops = {
>>  	.lookup = ptm_unix98_lookup,
>>  	.install = pty_unix98_install,
>> -	.remove = pty_unix98_remove,
>>  	.open = pty_open,
>>  	.close = pty_close,
>>  	.write = pty_write,
>> @@ -624,7 +619,6 @@ static const struct tty_operations ptm_u
>>  static const struct tty_operations pty_unix98_ops = {
>>  	.lookup = pts_unix98_lookup,
>>  	.install = pty_unix98_install,
>> -	.remove = pty_unix98_remove,
>>  	.open = pty_open,
>>  	.close = pty_close,
>>  	.write = pty_write,
> 
> thanks,

We can increment pty_count in open()
About BTW(->remove) You say in 24d406a6b:
However tty_shutdown() is called from queue_release_one_tty() only if
tty_operations->shutdown is NULL. But for pty, it is not.
pty_unix98_shutdown() is used there as ->shutdown.

So tty_operations->remove of pty (i.e. pty_unix98_remove()) is never
called. This results in invalid pty_count. I.e. what can be seen in
/proc/sys/kernel/pty/nr.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/